Mass 211 Analysis: the Big Picture

Current Landscape of Human Service Demands in Massachusetts

Jesse Yang

2017-03-12

Introduction

This report is a rough analysis of all the 211 calls made between Feb 25, 2016, and Feb 9, 20171 If I understood the data correctly, the new system is still in testing before May 2016, which was when calls started steadily coming in.. All the numbers and facts are for the whole time span, i.e., they are about the whole landscape of human service needs reached the Mass 211 system in a roughly 10 month period.

Key Findings

Drilling Down

The flow

Have you used Mass211 before? How did you hear of Mass211? AIRS I&R Need Category Referrals Made

Above graph shows where the calls were coming from and where the callers were referred to. The majority of calls are from first-time callers referred by human service agencies. Housing and Income Support/Assistance are the two most inquired issues. Most of Income Support/Assistance are about the Early Education and Care program (see below).

Calls with multiple purposes

About one-third of all calls had more than one Taxonomy term, and more than 47% callers were given multiple referrals.

Top 10 Level 5 Taxonomy term combinations

Categories Num. of Calls % of all calls
Electric Service Payment Assistance
  · Government Consumer Protection Agencies
258 0.83
Child Care Expense Assistance
  · Early Head Start Sites
189 0.61
Child Care Expense Assistance
  · Family Support Centers/Outreach * Children
142 0.46
Child Care Centers
  · Child Care Expense Assistance
124 0.40
Homeless Shelter
  · Housing Search Assistance * Tenants
124 0.40
Electric Service Payment Assistance
  · Utility Service Complaints
120 0.39
Child Care Expense Assistance
  · Head Start Sites
99 0.32
Electric Service Payment Assistance
  · Heating Fuel Payment Assistance
95 0.31
Child Care Expense Assistance Applications
  · Early Head Start Sites
  · Family Support Centers/Outreach * Children
85 0.28
Child Care Expense Assistance Applications
  · Family Support Centers/Outreach * Children
84 0.27

The most common case of multiple purpose calls were those asking for child care assistance programs, then information regarding child care providers (child care centers or early start sites) were also provided.

Another case is people making complaints about utility companies while seeking financial assistance in utility payment.

Calls with single purpose

There are many ‘useless’ calls in this category.

Top 10 Level 5 Taxonomy terms for single purpose call

Taxonomy Term Num. of Calls % of all calls
Child Care Expense Assistance 8116 26.26
Child Care Expense Assistance Applications 1233 3.99
Homeless Shelter 1029 3.33
211 Systems 1002 3.24
Rent Payment Assistance 989 3.20
Electric Service Payment Assistance 821 2.66
Directory Assistance 407 1.32
Food Pantries 275 0.89
Food Stamps/SNAP Applications 259 0.84
Housing Search Assistance * Tenants 225 0.73

Child care related calls took the largest portion of the caseload, in part because that Mass 211 has a unique collaboration with Massachussets Department of Early Education and Care (EEC),2 I have found no other states are having the same high-volume of calls related to child care. but also because some parents frequently called to check the status of their enrollment application for EEC subsidies.3 About 20% of child care related calls were explicitly marked as status check. One possible measure to reduce these calls is to better inform parents the expected processing time, if there exists a set expectation.

“211 Systems” are mostly border-line-crossing calls from neighboring states; “Directory Assistance” are those intended to call 411. They seem unavoidable.

Putting these categories aside, the most frequently inquired human services are EEC Applications, Homeless Shelter/Rent Assistance, Electricity, and Food.

Call length

Average call length of all calls is 8.3 minutes. Most calls (75%) finish within 10 minutes, 95% of them finish within 20 minutes. Some took as long as 40 minutes or more, but those were rare cases and often involved interpretation services (the caller didn’t speak English).

When removed undesired calls (211 Systems and Directory Assistance), which normally end fast, the average call length becomes 8.8 minutes.

Mean, median and quantiles of call length

minimum q1 median mean q3 maximum
0.5 3.5 6.5 8.334 10.5 403.5

If taking call length into consideration, the list of top categories by caseload would be different. Following are the average call length and number of calls aggregated to the Level 5 Taxonomy term.

The most time-consuming calls

Taxonomy Term Avg. call length Num of calls Minutes per day
Child Care Expense Assistance 7.83 10436 302.82
Rent Payment Assistance 9.95 5601 206.45
Homeless Shelter 10.95 4838 196.21
Electric Service Payment Assistance 9.43 4592 160.41
Child Care Expense Assistance Applications 13.28 2054 101.05
Food Pantries 11.77 1913 83.37
Heating Fuel Payment Assistance 11.71 1571 68.13
Food Stamps/SNAP Applications 10.92 1635 66.11
Housing Search Assistance * Tenants 10.44 1491 57.65
Family Support Centers/Outreach * Children 13.64 923 46.63

Note that for calls with multiple purposes, their call lengths are evenly divided for each purpose.

Use above interactive treemap to explore the average total time consumption per day for each category. Click on the tiles to enter a subcategory, then click on the title bar at the top to go back.

Geographic Distribution

Since most of the 211 services are about providing help to needy families and individuals, the geographic distribution of 211 calls strongly correlates with the income level of a region.

A multiple linear regression was calculated at the zip code level to predict the number of calls per 1,000 residents based on their income, race, and education level.

The variables chosen are percentages of people who completed at least a Bachelor’s degree (education_bachelor), percentages of white people (race_white), and median house income measured in 1,000 US dollars (income). An interaction term race_white:income is also added to reveal interactions between the effects of house income and race.

Dependent variable:
Number of calls per 1,000 people
Education: Bachelor and Above (%) -0.038***
(0.011)
Race: White (%) -0.184***
(0.018)
Median House Income (1K USD) -0.151***
(0.029)
White:MedianHouseIncome 0.001***
(0.0003)
Constant 22.628***
(1.388)
Observations 484
R2 0.509
Adjusted R2 0.505
Residual Std. Error 2.921 (df = 479)
F Statistic 124.012*** (df = 4; 479)
Note: *p<0.1; **p<0.05; ***p<0.01

A significant regression equation was found (F(4, 472) = 121.11, p < 0.000), with a \(R^2\) of 0.502.

The predicted number of calls per 1,000 people during a 10-month period is equal to 22.628 - 0.038 (education_bachelor) - 0.184 (race_white) - 0.151 (income) + 0.001 (race_white:income).

The number of calls would decrease 0.038 per 1 percent increase in people with a Bachelor’s degree, 0.184 per 1 percent increase in white residents, and 0.151 per $1,000 increase in median house income.

For different races (white and non-white), the effect of income may differ, but the difference is small (only 0.001)–at the same income level, white people is slightly more likely to call 211 than other races.

Concomitant requests

A call may naturally have multiple purposes, but does a call for certain service predict the coming of other types of needs?

To answer this question, we can check the concordance of need categories, i.e. the correlation between volumes of calls in different categories at a given region.

Take the most requested two service categories for example, zip code areas with higher demands for Income Support (Child Care Assistance) almost always have a high demand for Housing Assistance (Kendall’s tau coefficients \(r_{\tau}\) = 0.471, p < 0.000).

The same strong correlation could be found at most category combinations.

Concordance between service categories

Concordance between service categories

This graph shows Kendall’s tau coefficients for each combination of two AIRS Problem/Need categories. The darker the blue, the more in agreement the two categories are. Transparent tiles represent insignificant results, mostly because of low call volume. To minimize bias, when computing the coefficient for a pair of categories, zip codes with zero call in either of the categories were ignored.

The most obvious pairs are Art/Culture/Recreation + Clothing/Personal/Household Needs, Employment + Volunteers/Donations, Disaster Services + Volunteers/Donations, and Housing + Income Support.

It is not immediately clear whether calls of different categories are from the same group of people.

Notes

Data Treatment

This report is based on the iCarol reports generated on Feb 9, 2017. The data were manually cleaned and transformed, with a few noteworthy controls:

  1. Only regular Mass 211 calls are used4 ReportVersion = Mass 211. “Call2Talk”, “Runaway Form”, and “Mass 211 Text”, “Mass 211 Chat” were all ignored, the reason being that their number are small and they often contain incomplete or unique data fields, which means some separate analyses are more appropriate.
  2. Different data fields about the same information were merged into one. For example, “Data Collection - How did you Learn about Mass211” and “Caller Data - How Heard about Call2Talk?” are treated as one field.
  3. Some missing records of AIRS Problem/Needs categories were completed with inferences from the Taxonomy. E.g., “Target People -> Low Income” was mapped to “Income Support/Assistance”.
  4. All call length values were increased by 0.4 minutes, because they were zero-based and adding zeros up will introduce bias for the “minutes per day” metric.

Categorization

The Taxonomy provides a comprehensive and logical structure for human services, but is not suitable for high-level analyses–the end-level terms are too granular (there are 830 of them), and the upper levels too broad and not self-explanatory.

The AIRS: I&R Problem/Needs National Categories5 The linked document is outdated. It contains only 16 categories, but AIRS has split “Housing/Utility” into two separate categories in 2014, making it 17 categories in our case. as seen in the MetUnmet report is a better candidate for reporting, and was used in the flow chart at the beginning fo this document, but they are also not revealing enough. For instance, “Income Support” does not reveal the fact that the majority need is Early Education and Care.

That is why I used AIRS Neet Categories for general analaysis, but Level 5 Taxonomy terms to reveal more details in call purposes. Another solution is to create a flat, topic-based categorization method that aligns with the unique demands Massachusetts constituents. Curating such a list is a time-consuming process and would need vigorous validation.

Next Steps

We have completed the analysis of the big picture and created graphs and tools to explore basic characteristics of the demands.

The next step is to cross-validate with more data sources and start analyzing other data fields such as age, sex, and family characteristics.

But before doing that, building a simple and robust typology is yet still an unresolved challenge. I tried some name matching with Regular Expressions, but it did not work well for all cases.

However, if we are satisfied with current big picture presentation, we may also choose to dive into specific topics immediately. That is, to isolate calls related to a certain topic, for instance, people with mental health and substance abuse issues, then conduct comprehensive analysis in caller profiling, longitudinal trend, etc.

Questions

  1. Any comment in this big picture analysis? What else would you like to see?
  2. Are agency resources categorized in any way in the system? E.g., being tagged as Police, Municipality Service, Government Aide Program, Volunteer and Charity, Hospital, Shelters, etc. In the Flow diagram, I grouped the agencies by keywords in the names. Had we an official categorization method, the “Other Agency” chunk would be much smaller.
  3. Is the typology used by 211counts.org known to you? Did they develop the whole thing by themselves or is it yet another shared categorization system just like the Taxonomy and AIRS Problem/Need Categories?
  4. Did the merging of Call2Talk and 211 change the way you work or how you keep logs of these calls?
  5. Some Call2Talk calls were referred to Call2Talk (ReferralsMade = Call2Talk). Is it just redundant data, or does it have a special meaning?
  6. What’s the password of http://mass211.org/mass-211-stats/ ?